TSegFormer: 3D Tooth Segmentation in Intraoral Scans with Geometry Guided Transformer

This is the PyTorch implementation of TSegFormer [MICCAI 2023] [Paper link].

TSegFormer is a novel 3D tooth segmentation framework with a tailed 3D transformer and a multi-task learning paradigm, aiming at distinguishing the permanent teeth with divergent anatomical structures and noisy boundaries. Moreover, we design a geometry-guided loss based on a novel point curvature to refine boundaries in an end-to-end manner, avoiding time-consuming post-processing to reach clinically applicable segmentation.

Usage

Requirements

• python==3.7.11
• torch==1.9.0+cu111
• scikit-learn
• tqdm

Training nad testing

Put the IOS dataset in the ./data folder.

Run the training script for pretraining:

python main.py --epochs 200 --num_points 10000

The pre-trained model best_model.t7 is saved in ./outputs/exp/models.

Run the evaluation script with the pretrained model best_model.t7 for testing:

python main.py --eval True --model_path ./outputs/exp/models/best_model.t7

Citation

If you find our work useful in your research, please consider citing:

@article{xiong2023tsegformer,
      title={TSegFormer: 3D Tooth Segmentation in Intraoral Scans with Geometry Guided Transformer}, 
      author={Huimin Xiong and Kunle Li and Kaiyuan Tan and Yang Feng and Joey Tianyi Zhou and Jin Hao and Haochao Ying and Jian Wu and Zuozhu Liu},
      year={2023},
      eprint={2311.13234},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}